Artificial intelligence is no longer limited to transforming professional applications or digital creative environments; it is now becoming central to consumer behavior. During the presentation of its annual e-commerce results, Fevad unveiled a study conducted by Odoxa that highlights a significant shift: 31% of French online shoppers now use generative AI in their online shopping journey. This figure, particularly high given the still-recent maturity of these technologies, reveals a structural shift in interactions between consumers, platforms, and product information. AI is no longer content to simply optimize internal recommendations on e-commerce sites; it is becoming an active intermediary in the research, comparison, and sometimes even decision-making phases.
Rapid adoption driven by younger generations
The study shows that this adoption is not uniform but is strongly correlated with age, educational attainment, and digital literacy. Among 15–24-year-olds, nearly one in two young online shoppers—49%—uses AI for their online purchases. The rate remains high among 25–34-year-olds at 46%, as well as among executives at 44% and residents of the Île-de-France region at 40%. These figures reflect a marked generational and socio-professional effect, where proficiency with digital tools facilitates their integration into commercial practices.
This trend is all the more striking given the rapid growth in usage. Fifty-four percent of users report increasingly relying on AI for their purchases, suggesting a phenomenon of accelerated adoption. This trend is part of a broader movement toward the adoption of generative technologies in France: the proportion of French people who frequently use AI has risen from 6% in May 2023 to 25% today1. The arrival of ChatGPT and other conversational agents accessible to the general public has profoundly altered the speed of technological diffusion. Unlike previous waves of digital innovation, commercial adoption here immediately accompanies technological democratization.
AI is primarily used before a purchase is made
Agent-based commerce does not yet primarily occur at the time of the transaction itself, but rather beforehand, during the exploration and comparison phase. Fifty-eight percent of online shoppers who use AI do so before making a purchase, mainly to save time, synthesize information perceived as neutral, compare different offers, or narrow down their initial selection of products. AI thus acts as a cognitive filter, capable of condensing vast amounts of scattered information into structured recommendations.
In contrast, its use remains more limited at the time of purchase (27%) and after purchase (35%). This difference highlights a clear distinction between informational assistance and decision-making delegation. The level of trust measured in the study confirms this caution: 47% of respondents say they trust AI before the purchase, 43% after the purchase, but only 30% at the precise moment of the transaction. Payment is still perceived as an area of individual control, where algorithmic delegation meets with greater reluctance.
Complex technical products and services at the forefront
An analysis of the relevant categories shows that AI is used more frequently for purchases considered technical, time-consuming, or complex. Technical products and home appliances account for 29% of AI-assisted searches, even though they rank only 7th among the most frequently purchased categories online in 2025. This overrepresentation indicates that the use of AI is proportional to the perceived complexity of the decision.
Services also represent a prime area for agent-based commerce. Travel packages (23%) and transportation tickets (18%) are among the categories where AI is heavily utilized. These purchases involve multiple trade-offs—price, dates, options, and flexibility—and therefore particularly benefit from algorithmic assistance capable of comparing and anticipating. Conversely, fashion and apparel—despite being leaders in online sales—utilize AI in only 21% of cases. Decisions that are more intuitive or aesthetic in nature seem less suited to algorithmic mediation.
Regular users: a laboratory for agent-based commerce
The study highlights a clear divide between the general public and regular AI users. Today, 25% of French people use AI frequently, either daily or several times a week. Among them, 73% incorporate AI into their shopping journey, and 66% say they trust it in this context. These figures are significantly higher than those observed among online shoppers as a whole.
This correlation between frequency of use and trust illustrates a classic phenomenon of technology adoption, in which familiarity reduces the perception of risk and increases the perception of utility2. Reservations regarding commercial neutrality, for example, drop to 41% among regular users, compared to 57% in the general population. Agent-based commerce could therefore grow as the base of experienced users expands.
The purchase amount as a decisive factor
One of the study’s most significant findings concerns the role of price in agentic purchasing decisions. For 32% of respondents, price is the deciding factor. Purchases are more readily delegated to AI when the amount is less than 50 euros, suggesting that risk tolerance is proportional to the financial stakes.
The nature of the product or service is also a factor, cited by 30% of respondents. Whether or not a person is already a customer of the e-commerce site plays a more marginal role (19%), indicating that trust in the platform does not entirely dispel concerns about AI itself. Among regular users, the idea of fully delegating payment to AI is more widely considered—27% versus 22% in the general population—which could signal the gradual emergence of semi-autonomous transactions.
Legal, Ethical, and Economic Issues
The main concerns identified relate to the commercial neutrality of AI (57%), data privacy (51%), and payment security (46%). These concerns are directly linked to the requirements of the European Artificial Intelligence Regulation adopted in 2024, which imposes stricter obligations regarding transparency, documentation, and risk management for general-purpose AI systems3.
The growth of agent-based commerce will therefore depend on several interrelated factors: technical robustness, payment security, regulatory compliance, and consumer education. This is less of an obstacle to e-commerce—which is already well established—than it is a specific barrier to the acceptance of AI itself.
A shift in consumer behavior that is already underway
In less than three years, artificial intelligence has become integrated into French consumer habits at an unprecedented pace. The fact that nearly one in three online shoppers already uses it is a strong indicator of early maturity. Moreover, commercial adoption is keeping pace with technological diffusion, a phenomenon rarely seen at this stage of adoption.
The future of agent-driven commerce will now depend on how trust evolves at the critical moment of the transaction. As the number of regular users grows, reservations may fade, paving the way for forms of partially delegated purchasing. The question is no longer whether AI will have a lasting impact on e-commerce, but how it will reshape the relationship between consumers, platforms, and decision-making.
Learn more
The rise of artificial intelligence in the shopping experience isn’t limited to individual use; it’s also reshaping the strategies of major platforms. On a related topic, check out our article “Amazon vs. Perplexity: The Battle for the Future of Online Commerce”, which analyzes how e-commerce players are integrating generative AI and conversational engines to transform product search, recommendations, and the customer experience.
References
1. EAB. (2026). How Students View and Use AI in College Search. How Students View and Use AI in College Search | EAB.
https://eab.com
2. Bubeck, S. et al. (2023). Sparks of Artificial General Intelligence. Microsoft Research.
https://www.microsoft.com/en-us/research
3. OECD. (2023). AI Compute and Industrial Scaling.
https://oecd.org
4. World Economic Forum. (2023). The Future of Jobs Report.
https://www.weforum.org
5. European Parliament. (2024). Artificial Intelligence Act.
https://www.europarl.europa.eu
